Tube-nosed variations–a new species of the genus Murina (Chiroptera: Vespertilionidae) from China

During a survey in 2014, several tube-nosed bats (Vespertilionidae: Murininae: Murina) were collected in Sichuan Province. Based on morphological characters, these bats did not match any species previously recorded from China. Morphometric analyses and phylogenetic inference based on mitochondrial and nuclear gene sequences indicated that they represented a new species, named here as Murina jinchui sp. nov. Although the new species is presently known only from Wolong National Nature Reserve, it is unlikely to be a rare species in the area based on our capture frequencies.


DEAR EDITOR,
During a survey in 2014, several tube-nosed bats (Vespertilionidae: Murininae: Murina) were collected in Sichuan Province. Based on morphological characters, these bats did not match any species previously recorded from China. Morphometric analyses and phylogenetic inference based on mitochondrial and nuclear gene sequences indicated that they represented a new species, named here as Murina jinchui sp. nov. Although the new species is presently known only from Wolong National Nature Reserve, it is unlikely to be a rare species in the area based on our capture frequencies.
In 2014, several small-sized and impressively colored Murina individuals were collected during field surveys in Sichuan Province (all field surveys and sample collection protocols complied with the current laws of Sichuan Province). Morphological and molecular biological examinations revealed them to be distinct from all other recognized Murina taxa; therefore, they are described herein as a new species.
For morphometric analysis, we examined 224 specimens of Murina deposited in 12 collections (see list of specimens in Supplementary Material (Appendix I)). Six external measurements (to the nearest 0.1 mm), body mass (to the nearest 0.1 g), and 16 craniodental measurements (to the nearest 0.01 mm) were recorded by the same author. Definitions and details of measurements are listed in the Table  1 and Supplementary Material (Methods).
We performed principal component analysis (PCA) and discriminant analysis of principal components (DAPC) (Jombart, 2008;Jombart et al., 2010) for species discrimination based on external and craniodental measurements. For both PCA and DAPC, the sexes were analysed separately because sexual dimorphism is noted in  Figure S1 and Table S1). We also replicated our multivariate statistical analyses in the monophyletic clade formed by M. leucogaster, M. shuipuensis, M. rongjiangensis, and the new species. In addition, due to their similarities in size and skull proportions, M. shuipuensis, M. rongjiangensis, and the new species were also compared using analysis of variance (ANOVA). All analyses were performed using the "phych" and "adegenet" packages in R (Jombart, 2008; R Development Core Team, Vienna, www.R-project.org). For phylogenetic analysis, the partial cytochrome oxidase subunit I (COI, 670 bp) and partial nuclear recombination activating protein 2 genes (Rag2, 1 339 bp) were selected as molecular markers (Heaney et al., 2012;Kuo et al., 2017;Roehrs et al., 2010). The COI gene was amplified from specimens of all species in this study, whereas the Rag2 gene was sequenced from single individuals of each species confirmed by our phylogenetic and morphological species determination (GenBank accession Nos.: MN549027-MN549101). All available COI and Rag2 sequences of Murina collected from NCBI-nt and our specimens were aligned using MUSCLE (Edgar, 2004). Final alignment was partitioned by different codon positions and the parameters of the best nucleotide substitution models were determined by PartitionFinder2 (Lanfear et al., 2017) using the greedy algorithm (Lanfear et al., 2012). Maximum-likelihood (ML) trees were searched in RAxML v7.4.2 (Stamatakis et al., 2008), and the reliability of nodes was evaluated by 500 rapid bootstrap matrixes.
To describe pelage color, digital photographs of freshly euthanized bats were taken in the field (see details in Supplementary Material). Following Davis & Castleberry (2010), images were color calibrated, and pelage color was described following the Pantone color code with the redgreen-blue (RGB) system.
Detailed measurements of all specimens used in this study are listed in Table 1. The PCA based on six external measurements revealed that 76% of total variance could be explained by the first two components (54% and 22% for principal components 1 (PC1) and principal components 2 (PC2), respectively) ( Figure 1A). For PC1, all measurements, except for ear length, had high positive loadings, thus reflecting an external overall size effect. PC2 was mostly related to ear size (larger value indicates larger measurement) ( Figure 1A, with values provided in Table 2). Although the interspecific patterns revealed by PCA were ambiguous, implying difficulty in species identification based on external measurements only, the PCA results consistently indicated larger females and smaller males within Murina species. For PCA of the 16 craniodental measurements, 81% of total variance could be explained by the first two components (57% and 24% for PC1 and PC2, respectively) ( Figure 1B). For PC1, 11 measurements had positive loadings ( Figure 1B, with values provided in Table 1), suggesting that this PC was mainly related to skull size (smaller bats characterized by lower PC1 scores). Thus, with the exception of M. leucogaster, most Murina species with suilla-type dentition (crown area of upper canine less than that of P 4 ) had smaller skulls than those species with cyclotis-type dental characters (crown area of upper canine equal to or larger than that of P 4 ) ( Figure 1B). For PC2, most measurements had low loadings, except for braincase width (BCW), mastoid width (MAW), and interorbital width (IOW) ( Figure 1B and Table 1). Unlike the ambiguous pattern found for external measurements, PCA based on craniodental measurements revealed a more  (1) Values are given as Means±SD (if n>5), and minimum-maximum (min -max). t-value is from Students t-test between sexes when distribution of measurement fits normality, and * and NS represent P<0.05 and non-significant result, respectively. Abbreviations and definitions for measurements are as followed: HB: Total length-from the tip of the face/chin to the anus; T: Tail  noticeable interspecific relationship, with six skull-size assemblages identified: i.e., (1) M. leucogaster; (2) M. aurata; (3) all taxa with cyclotis-type dentition included in this study; (4) M. eleryi+M. chrysochaetes; (5) M. shuipuensis+M. rongjiangensis+new species; and (6) "M. jaintiana+M. beelzebub+M. feae" (Figure 1B).
We used the PC1 and PC2 values from the external and craniodental analyses for DAPC, with the results together explaining 87% and 90% of total variance, respectively. Among the original external variables, HB, T, and FA had high positive loadings on discriminant function 1 (DF1), whereas HB and T had high positive loadings on discriminant function 2 (DF2) ( Table 2). For craniodental DAPC analysis, greatest length of skull (GTL), condylocanine length (CCL), and greatest length of mandible (ML) contributed substantially to DF1, whereas braincase height (BCH) and mastoid width (MAW) contributed substantially to DF2 (Table 2). Similar to the PCA plots of external variables, the DAPC plots revealed an ambiguous pattern, with most taxa overlapping ( Figure 1C).
Nevertheless, based on an increase in discriminant power, the pattern of the craniodental DAPC plots was more distinguishable than that of the PCA plots, although most were still partially overlaid ( Figure 1D). Similar to the PCA pattern from craniodental measurements, a noticeable interspecific relationship emerged, with five skull-size assemblages  represent clusters of multiple specimens, with horizontal dimension proportional to amount of sequence divergence. Solid triangles indicate newly generated COI sequences. Numbers above branches indicate level of bootstrap support >50% for the branch. power in determining species and sex than the external one, nevertheless both received an ambiguous reclassification in determining M. shuipuensis, M. rongjiangensis, and our new species and their sex (PP<0.80 for the most likely group and its actual allocation) ( Figure 1E-F). These results indicate external and craniodental similarity as well as difficulty in identification using morphological or craniodental measurements alone.
COI alignment spanned 671 bp, including 265 and 256 variable sites and parsimony informative sites, respectively. Rag2 alignment covered 1 339 bp, including 127 variable sites and 58 parsimony informative sites. For both alignments, the best partitioning scheme selected was one separating each codon position into a partition. For ML analyses, the best nucleotide substitution models for the first, second, and third partitions of COI were TIM+G, HKY, and GTR+G, respectively, whereas the best nucleotide substitution models for three position of Rag2 were TRN+I, HKY, and TIM+G, respectively. The ML trees recovered the genus Murina as a well-supported (BS=100) monophyletic group ( Figure 1G -H). The phylogenetic reconstructions of the COI sequences revealed similar species groups as reported by Francis & Eger (2012) and Eger & Lim (2011). Within the genus, a robust clade, including several Murina species known from southern China, emerged (highlighted with gray rectangle in Figure 1G, BS=100). The new species formed a basal, well-supported, monophyletic group within this clade, which included M. rongjiangensis, M. fanjingshanensis, M. shuipuensis, M. hilgendorfi, and M. leucogaster. Due to the limited number of phylogenetically informative variations, phylogeny using Rag2 resulted in a more ambiguous topology ( Figure 1H). However, a well-supported clade (highlighted with gray rectangle in Figure 1H, BS=82), similar to the species assemblage based on the COI marker, also emerged, with the basal position of the new species supported ( Figure 1GH). Hence, the distinctiveness of the new species was supported by both mitochondrial and nuclear markers.
X Skull small and delicate ( Figure 2F; Table 1; Supplementary  Table S1), with rostrum sloping gently to forehead ( Figure 2F). Laterally, mid-portion of braincase exceeds frontal region in height, braincase not especially domed. No sagittal crest, lambdoid crests relatively weak ( Figure 2F). Rostrum deep and pronounced. Depth of nasal emargination exceeds its width, approximately extending to middle of upper canine, outline of emargination varying from smoothly concave to squarish in dorsal view. Basisphenoid pits well-defined, teardrop shaped, elongated, and deep. Zygoma weak, lacking dorsal processes ( Figure 2F). Postpalatal emargination with medial projection. Dental formula I 2/3 C 1/1 P 2/2 M 3/3 ( Figure 2G). Maxillary toothrows convergent anteriorly (RCM: =0.69,SD=0.02,n=6). I 2 bicuspid, situated anterior to I 3 ; I 3 with small secondary cusp lingual to primary cusp ( Figure 2F). I 2 and I 3 equal in height, first about half of basal area of second upper incisor; second upper incisor in contact with upper canine, approximately half height of C. Basal area of canine equal to that of first premolar, appearing circular in cross-section. P 2 delicate and pointed, about half P 4 in height; P 4 distinctly higher than C, with basal area twice as large. Metacones of M 1 and M 2 exceeding respective paracones in height. Mesostyle of M 1 and M 2 moderately developed but retaining distinct cusp ( Figure 2G). Posterior upper molar lacking metacone, with reduced but distinct postparacrista. Mandible delicate, and lower incisors tricuspid. Lower canine exceeding posterior two premolars in height and basal area, well-developed inner cingulum with anterior-most portion in contact with posterior face of I 3 . Basal area of P 2 approximately two-thirds that of P 4 , height of P 2 approximately equal to that of P 4 . First and second lower molars nyctalodont, with well-developed entoconids. Talonids of M 1 and M 2 equal to respective trigonids in crown area, M 3 talonid half of trigonid ( Figure 2F, G). Comparisons with other taxa: Murina jinchui sp. nov. possesses suilla-type dentition having an upper canine smaller basally than the corresponding posterior premolar, and thus is readily distinguished from all species with cyclotistype dental characters. Among all recognized species with suilla-type dentition of the phylogenetically most closely related species, M. hilgendorfi, M. leucogaster, M. bicolor, and M. fanjingshanensis all are much larger (FA over 37 mm, Table 1); the similar-sized M. shuipuensis and M. rongjiangensis have an overall reddish dorsal fur, distally yellowish-gold ventral hairs, shorter ears, shorter FA, less convergent maxillary toothrow, and weaker P 4 (Table 1; Supplementary Figure S1 and Table S1).
Other species with a similarly developed P 4 , which is basally much larger than C and P 2 (M.  dimensions. Etymology: The species is named in honor of Professor Jinchu Hu, one of the founders of the Hetaoping Giant Panda Training Base and Wolong National Nature Reserve in China as well as a well-known giant panda expert, who has supported, encouraged, and actively participated in extensive ecological research in China, especially in Sichuan Province. Natural history: Currently known only from the type locality, namely Hetaoping Giant Panda Training Base, Wolong National Nature Reserve, Wenchuan County, Sichuan Province, China. The Murina jinchui sp. nov. specimens were captured in harp traps set in moist evergreen and deciduous broadleaved forests (e.g., Cyclobalanopsis, Quercus, Betula) near the ruins of the original Giant Panda Breeding Center of the Hetaoping Giant Panda Training Base. Three Rhinolophus ferrumequinum and one Pipistrellus sp. were also caught at this site.
The existence of cryptic diversity within Murina is demonstrated by the large number of new species described from the Indomalayan region. China's wide variety of habitats, ranging from tropical to boreal forests and from grassland to desert, has greatly contributed to the richness of chiropteran resources. Murina species from China are recorded from temperate regions in the north to subtropical and tropical regions in the south and southwest (Jiang et al., 2015;Liu & Wu, 2019). This vast distribution, together with the many unexplored areas in regard to chiropterological surveys, implies an exceptionally high level of cryptic species diversity in tube-nose bats. Although molecular analysis using DNA barcoding, developed as a tool for rapid identification, can facilitate cryptic species recognition and improve understanding of biogeographical patterns (Csorba et al., 2011;Francis et al., 2010;Francis & Eger, 2012), taxonomic and systematic studies and species identification still depend on morphological data. For instance, species identification of Murina typically requires evaluation of morphological characters and availability of properly identified comparative material in museum collections, most of which is deposited in Europe and North America. Such predicaments may be abated in the future by reciprocation of 3D skull models of comparative material using high precision 3D scanners.
Mammalian fur/pelage within many species is subject to a considerable degree of intraspecific color variation (Corbet & Hill, 1992;Davis & Castleberry, 2010;Kries et al., 2018;Son et al., 2015); however, the color patterns of Murina, including overall dorsal and ventral aspects, banding of individual hairs, and presence and distribution of brightly colored guard hairs on forearm and/or hindfoot, appear to be stable within species. Therefore, coloration is a useable diagnostic tool for identifying Murina species in the field.
Although no sexual or intraspecific variations in pelage color were observed in the new species, sexual size dimorphism, a common phenomenon in Murina (Son et al., 2015), emerged during morphological analyses ( Figure 1A-C, E; Table 1; Supplementary Figure S1). Females were significantly larger than males in measurements related to length and width of braincase, upper toothrow, and overall size of lower jaw (Table 1). This pattern of sexual dimorphism is suggested to have occurred under functional limitations and differed among skull regions (e.g., braincase and nasal chambers), therefore doesn't reflect simple allometric size changes.

SUPPLEMENTARY DATA
Supplementary data to this article can be found online.